基于图像亮度局部约束的总变分图像去噪法  

Total Variation Image Denoising Based on Local Intensity Bound Constraints

作　　者：陈永彬[1] 宋海英[1] 

机构地区：[1]成都电子机械高等专科学校通信工程系,成都610031

出　　处：《数据采集与处理》2011年第5期564-568,共5页Journal of Data Acquisition and Processing

摘　　要：为了解决总变分图像去噪方法易产生虚假边缘这一不足,结合总变分理论和图像亮度局部约束的优点,提出了一种新的图像去噪方法。在总变分图像去噪的数值求解过程中,利用局部均值和局部方差对迭代过程中所获得的结果进行局部约束。实验结果表明,本文方法具有良好的图像去噪能力,在高噪声水平下,其峰值信噪比(Peak signal-noise ratio,PSNR)至少提高了1dB左右,图像视觉效果也有了很大改善。A new image denoising method is proposed, which combines the total variation theory with the local intensity bound constraints. To overcome the weakness that the total variation image denosing method is easy to produce false edges, local mean and local variance are utilized to constrain the numerical solution at each iteration step. Experiments show that the proposed method is improved in terms of peak signal-noise ratio （PSNR） and visual effect. Especially when the noise level is high, PSNR is improved by at least 1.0 dB compared with the total variation denosing method.

关 键 词：图像去噪 总变分 局部约束 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]